Abstract:
Healthcare is approaching one-fifth of the U.S. economy, yet with inequitable access, first-in-human-history demographic shift to older age, and unsustainably inflated per capita costs; this gathering storm underscores the urgent need for innovative solutions. In this seminar, I will share how to train a visual-language foundation model for pathology using publicly available medical images, and how that publicly shared medical information can be harnessed to enhance diagnosis, knowledge sharing and education. I will share nuclei.io, the state-of-the-art active learning software for digital pathology, and the benefit of human-AI collaboration across two diagnosis tasks. Furthermore, I will demonstrate how to leverage AI and computational approach to forecast post-chemotherapy outcome from pre-treatment tissue biopsies, thus enhance the treatment strategies. Finally, I will outline my future research directions, and how to train medical AI in the same way as humans.

Bio:
Dr. Zhi Huang is a postdoctoral fellow at Stanford University. He received his Ph.D. degree from Purdue University, in August 2021, majoring in Electrical and Computer Engineering (ECE). His research focuses on Artificial Intelligence, Human-AI Collaboration, Medical Image Analysis, and Precision Medicine.

Zoom:
https://umich.zoom.us/j/94801149707